trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.006839945280437756,
  "eval_steps": 500,
  "global_step": 10,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 3e-05,
      "loss": 0.8018,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.046106882113753e-07,
      "loss": 0.8749,
      "step": 10
    },
    {
      "epoch": 0.01,
      "step": 10,
      "total_flos": 835654992789504.0,
      "train_loss": 0.8675599694252014,
      "train_runtime": 15.492,
      "train_samples_per_second": 2.582,
      "train_steps_per_second": 0.645
    }
  ],
  "logging_steps": 10,
  "max_steps": 10,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 835654992789504.0,
  "trial_name": null,
  "trial_params": null
}